Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
DISCUSSION PAPER SERIES
IZA DP No. 11470
Rania GihlebOsea GiuntellaNing Zhang
The Effects of Mandatory Prescription Drug Monitoring Programs on Foster Care Admissions
APRIL 2018
Any opinions expressed in this paper are those of the author(s) and not those of IZA. Research published in this series may include views on policy, but IZA takes no institutional policy positions. The IZA research network is committed to the IZA Guiding Principles of Research Integrity.The IZA Institute of Labor Economics is an independent economic research institute that conducts research in labor economics and offers evidence-based policy advice on labor market issues. Supported by the Deutsche Post Foundation, IZA runs the world’s largest network of economists, whose research aims to provide answers to the global labor market challenges of our time. Our key objective is to build bridges between academic research, policymakers and society.IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion. Citation of such a paper should account for its provisional character. A revised version may be available directly from the author.
Schaumburg-Lippe-Straße 5–953113 Bonn, Germany
Phone: +49-228-3894-0Email: [email protected] www.iza.org
IZA – Institute of Labor Economics
DISCUSSION PAPER SERIES
IZA DP No. 11470
The Effects of Mandatory Prescription Drug Monitoring Programs on Foster Care Admissions
APRIL 2018
Rania GihlebUniversity of Pittsburgh and IZA
Osea GiuntellaUniversity of Pittsburgh and IZA
Ning ZhangUniversity of Pittsburgh
ABSTRACT
IZA DP No. 11470 APRIL 2018
The Effects of Mandatory Prescription Drug Monitoring Programs on Foster Care Admissions*
The opioid epidemic is a national public health emergency. As the number of fa- tal
overdoses and drug abuse skyrockets, children of opioid-dependent parents are at increased
risk of being neglected, abused or orphaned. While some studies have examined the effects
of policies introduced by states to restrict prescription drug supply on drug abuse, there is
no study analyzing their effects on children. This paper estimates the effect of must-access
prescription drug monitoring programs (PDMPs) on child removals. To identify the effects of
the programs on foster care caseloads, we exploit the variation across states in the timing
of adoption of must-access PDMPs using an event-study approach as well as standard
difference-in-difference models. Consistent with previous evidence examining the effects of
PDMPs on drug abuse, we find that operational PDMP did not have any significant effects
on foster care caseloads. However, the introduction of mandatory provisions reduced child
removals by 10%. Exploring the reasons of removals, we show that these effects are driven
by the reductions in cases of child neglect. There is also evidence of significant reductions
in removal cases associated with child physical abuse.
JEL Classification: I12, I18, J13
Keywords: opioid epidemic
Corresponding author:Rania GihlebDepartment of EconomicsUniversity of Pittsburgh4901 Wesley W. Posvar Hall230 South Bouquet StreetPittsburgh, PA 15260USA
E-mail: [email protected]
* We are grateful to all the seminar attendees at the University of Pittsburgh. We benefitted from comments from
Janet Currie, Mark Borgshulte, Jason Cook, and Betsey Stevenson. We are thankful to the National Data Archive on
Child Abuse and Neglect for providing the data on foster care caseloads and child maltreatment.
1 Introduction
The United States are in the midst of an opioid overdose epidemic. Drug mortality rose
by 300% between 1999 and 2016. In 2016, the US experienced the largest annual increase
in drug overdose deaths ever recorded. This rapid increase in drug mortality is related to
the diffusion of prescription opioids (e.g., Oxycontin) and more recently to the spread of
fentanyl, an opioid typically used as a pain medication (Paulozzi et al., 2014; Dart et al.,
2015; Ruhm, 2018). Public health officials consider the current opioid epidemic crisis the
worst drug crisis in America history.
As recently highlighted by Quast et al. (2018), a critical aspect of this drug crisis are its
effects on the ability of addicted parents to care for their children. In 2015, there were 683,000
victims of child abuse and neglect reported to child protective services (CPS). According to
the US Department of Health and Human Services (2015), 25.4% of victims of child abuse
were reported with the drug abuse caregiver risk factor and the evidence suggests there
is an increase in caregiver drug abuse. Parental neglect and parental drug abuse are the
two most common reasons for removals (AFCARS Report, 2015). The opioid epidemic has
forced thousands of children from their homes, at risk of being neglected, abandoned, or
orphaned by drug-addicted parents. There is increasing concern that America’s opioid crisis
may overwhelm the US foster case system as thousands of children are taken out of the care
of addicted parents.1 Foster care numbers have been soaring in many US states. In 2015,
429,000 children were in foster care with a 6.7% increase with respect to 2013. This surge
is in large part due to the increase in the number of cases related to parental drug-abuse.
Figure 1 illustrates how the trend in drug-related child abuses and the number of children in
foster because of drug-related abuses closely mirrored the increase in fatal overdoses. From
2000 to 2015 number of drug-related foster care caseloads have increased by 66% (Figure
2). These increases in the foster care population generate significant monetary and non-
1See https://www.npr.org/2017/12/23/573021632/the-foster-care-system-is-flooded-with-children-of-the-opioid-epidemic
2
monetary costs. Previous research estimate that the fiscal costs of a child in foster care is
approximately $20,000 (Zill, 2011) and documents that foster care placement can have large
detrimental effects on children long-term outcomes (Doyle Jr, 2007, 2008). Of course this is
only a part of the costs as child abuse and neglect have long-run effects on human capital,
health outcomes and have been shown to increase the likelihood of engaging in crime and of
substance abuse (Currie and Spatz Widom, 2010; Dube et al., 2003).
To address the surge in prescription drug abuse, states have adopted prescription drug
monitoring programs (PDMP). These programs track prescriptions helping in identifying
doctor shopping and prescription drug abuse. Currently, most states have adopted an op-
erational PDMP, but only a few states mandated their use. The main contribution of this
study is to evaluate the effects of prescription drug monitoring programs on child removals.
To the best of our knowledge, this is the first study analyzing the effects of prescription drug
monitoring programs on foster care admissions.
Despite the growing attention raised by the press reports of state foster systems being
overwhelmed by children of opioid-dependents, there is little empirical evidence documenting
the relationship between opioid abuse and child removals. Cunningham and Finlay (2013)
analyzed the effect of meth use on foster care admissions using an instrumental variable
strategy. Their strategy relies on deviations in the real price of meth from national trends
caused by large federal supply interdictions that affected meth supply. They find evidence
of a positive elasticity of foster care with respect to meth use. However, a limitation of their
empirical strategy is that it only exploits national variation in meth prices and thus cannot
control for unobserved time-varying factors that may have caused changes in foster care
caseloads. Using data from Florida counties for the period 2012-2015, Quast et al. (2018)
document that an increase in opioid prescription rate was associated in an increase in the
removal rate for parental neglect, with the effects largely driven by counties with the highest
concentration of whites. Their analysis provides insightful findings, however it is limited
by the short sample period, the local nature of the data which may not be representative
3
of the US population and the small sample size which restricts their ability to control for
county-level time-varying confounding factors.
To identify the effects of PDMPs on foster care admissions we exploit variation in the
timing of adoption of operational and mandatory PDMPs across US States using an event-
study as well as standard difference-in-difference regression models. Consistent with previous
studies analyzing the effects of PDMPs on drug abuse (Buchmueller and Carey, 2017; Dave
et al., 2017), we find no evidence that operational PDMPs had significant effects on foster
care admissions, while mandatory-access PDMPs reduced child removals by 10%. Our results
suggest that mandatory PDMPs may reduce foster care costs by 476 million per year. Given
the long-lasting implications of child maltreatment and neglect, our results suggest that
programs aimed at controlling the supply of prescription drugs may have large long-run
returns if effectively enforced.
The paper is organized as follows. Section 2 discusses the background and presents the
main data sources. We discuss the empirical specification in Section 3. Results are presented
in Section 4. Section 5 concludes.
2 Background and Data
2.1 Prescription Drug Epidemic and Policy Response
According to CDC estimates (Rudd, 2016), the rise in prescription drug use and abuse
largely accounts for the trends in drug-related deaths. While the reasons behind the opioid
epidemic are multiple have also been linkeds to long-run socio-economic decline (Case and
Deaton, 2015), a growing set of studies relates the opioid epidemic to physician behavior
and supply-side regulation (Alpert et al., 2017; Pacula et al., 2015; Ruhm, 2018). Among
other factors, the market entry of OxyContin in 1996 and the diffusion of aggressive pain
management contributed substantially to the surge in opioid use over he last two decades
(Laxmaiah Manchikanti et al., 2012). Reports also suggest that most of the individuals at
4
high risk of fatal overdose obtained prescription drugs from physicians and doctor shopping
is considered the main source of supply.
To respond to the dramatic increase in fatal overdoses and drug-abuse, states have intro-
duced several programs to improve opioid prescribing, inform clinical practice and protect
patients at risk. Prescription drug monitoring programs (PDMPs) are electronic databases
that track controlled substance prescriptions in a state. PDMPs allow health authorities and
pharmacies to have timely information about prescribing and patient behaviors and can help
identifying patients who are receiving multiple prescriptions and may contribute to the epi-
demic. Some states made the use of the database mandatory for physicians. Non-mandated
PDMPs do not legally require health professionals to query them. However, since 2007 a
few states have know extended their PDMP with mandatory access provisions which require
doctors and pharmacies to query PDMP before prescribing a controlled substance.
Previous studies evaluating the effects of PDMPs on opioid consumption reached different
conclusions. There is consensus that PDMPs reduced oxycodone shipments (Kilby, 2015;
Mallatt, 2017). The evidence is mixed when focusing on hydrocodone shipments or other
abuse outcomes. While some studies found evidence that non-mandated PDMPs decreased
fatal non-oxycodone related overdoses and poisonings (Mallatt, 2017; Patrick et al., 2016;
Simoni-Wastila and Qian, 2012), most of them found evidence of small or null effects drug
abuse (Simoni-Wastila and Qian, 2012; Meara et al., 2016). On the contrary, recent papers
focusing on the effects of PDMP mandates found significant effects on opioid quantity and
shopping behavior, abuse outcomes, substance abuse facility admissions, crime rates, and
fatal drug overdoses (Buchmueller and Carey, 2017; Patrick et al., 2016; Dave et al., 2017;
Borgschulte et al., forth.; Mallatt, 2017).
To the best of our knowledge there is no study analyzing the effects of PDMP programs
on foster care caseloads. In our analysis, we distinguish the effects of operational PDMPs
from the effects of program that introduced mandatory access.
5
2.2 Data
Data on foster-care cases are drawn from the Adotpion and Foster Care Analysis and Re-
porting System (2000-2015). The Adoption and Foster Care Analysis and Reporting System
(AFCARS) is a federally mandated data collection system providing case specific informa-
tion on all children covered by the protections of Title IV-B/E of the Social Security Act
(Section 427). The foster care data files contain information on child demographics including
gender, birth date, race, and ethnicity.We calculated the number of foster-care cases by year
and state.
In addition, we collected county level controls from the Population and Housing Unit
Estimates (PHUE, 2000-2015), the US Census (2000) and the American Community Survey
(2001-2015). We use data from the PHUE to calculate child population and the share of
children who were victims of child maltreatment. We include data on age composition,
share of African-american population, share of Hispanic population, median income, gender
composition, and unemployment rate drawn from the 2000 US Census and the 2001-2015
American Community Survey. Finally, we collected data on the timing of adoption of other
laws that may have affected prescription drug abuse (e.g., Good Samaritan laws, Doctor
Shopping, Pain Clinic regulations, Physician exams laws, require ID laws, and tamper-
resistant prescription form requirement laws).
3 Empirical Specification
To identify the dynamic response of foster care caseloads to drug monitoring programs we
employ an event-study methodology and estimate the following equation:
Childst = δs + φt +4∑−4
γtMandates,t−τ +Xst + πs(δs ∗ t) + �st (1)
6
where Childst is the number of new foster-care admission in year t in state s. Given the
skewed distribution of foster care caseloads we use the inverse hyperbolic transformation
(IHS) of foster cases as our dependent variable.2. Mandatest is an indicator for whether
state i has introduced a mandatory PDMP in year t. Xst are a set of time-variant state
level controls (age composition, share of African-american population, share of Hispanic
population, median income, gender composition, and unemployment rate). All our estimates
control for the natural logarithm of the child population (aged 0-18). δs are state fixed effects
that capture time-invariant state level characteristics; φt are year fixed effects capturing the
average national trend in child abuse; and δs ∗ t are state specific time trends. Standard
errors are clustered at the state level. All estimates are weighted by child population.
Dave et al. (2017) show that adopters of mandatory access provisions were very similar to
states having a PDMP but who did not adopt such provisions. For this reason, to investigate
the role of Mandatory access PDMPs, we restrict the analysis to states with an operational
PDMP. The underlying assumption is that the states that had an operational PDMP but
did not introduce a mandatory access provision provide a valid counterfactual for treated
states (states with mandatory access provisions).
We then investigate the effects of PDMPs and mandatory PDMPs in a standard differences-
in-difference specification. Our identification strategy relies on the assumption that prior to
the adoption of drug monitoring programs treated and untreated states were following par-
allel trends and in the absence of programs implementation their path would have not been
affected. In particular, we identify the effects of the program exploiting within-state changes
in trends at the timing of implementation of drug monitoring programs. Consistent with
the evidence from previous work on the effects of PDMPs on drug abuse (Buchmueller and
Carey, 2017; Dave et al., 2017) and based on the dynamic response found in our event study
that show the effects of mandatory PDMPs materializes two years after the enactment, we
use a two-year lag to estimate our difference-in-difference model. Lagged effects are explained
2In the Appendix, we show that results tend in the same direction when using the number of cases orthe number of cases per 1,000 individuals as alternative scales
7
by the fact that it takes time for provider practices to diffuse across the state and there is a
natural lag between increased prescription drug monitoring and the reduction in the overall
supply of drugs.
In practice, we estimate the following OLS model
Childit = α + βMandatei,t−2 + ψXit + si + γt + �it (2)
where Childit is the number of abuses or foster-care admission in year (or month) t
in state i. Mandatei,t+2 is an indicator for whether state i has introduced a must-access
Prescription Drug Monitoring Program by year t. Xit are a set of time-variant state level
controls (age composition, share of African-American population, share of Hispanic pop-
ulation, median income, gender composition, and unemployment rate). All our estimates
control for the natural logarithm of the child population (aged 0-18). In addition, we control
for the adoption of other laws that may have affected prescription drug abuse (e.g., Good
Samaritan laws, Doctor Shopping, Pain Clinic regulations, Physician exams laws, require
ID laws, and tamper-resistant prescription form requirement laws). Finally, we include year
fixed effects, capturing the average national trend in child abuse, state fixed effects that cap-
ture time-invariant county level characteristics, and state specific time trends. All estimates
are weighted by child population.
4 Main Results
4.1 Trends and Descriptive Statistics
Over the last few years there has been a sharp increase in the number of child removals.
This trend closely mirrors the increase in fatal overdoses (Figure 1) and is largely driven by
the increase in the cases associated with child neglect or caregiver drug-abuse (Figure 2).
These figures parallel the dramatic surge in the distribution of prescription drugs across the
8
US (Figure 3).
Table 1 provides summary statistics of our main outcomes of interest. There were on
average approximately 4,500 child removals in a state-year. The two main reason for removals
are parental neglect and parental drug abuse. In 70% of the cases, removals were associated
with parental neglect, while in 31% of the removals were associated with parental drug abuse.
The share of removals associated with parental drug abuse rose from 22% in 2000 to 39% in
2015.
4.2 PDMPs and Foster Care Caseloads
In Figures 4-5, we explore the dynamic response of child removals to the adoption of op-
erational and then mandatory PDMPs. The event-study analysis shows that there was no
significant effect of PDMP on child removals (Figure 4). On the contrary, following the adop-
tion of must-access PDMPs we observe a marked decline in the number of child removals
(Figure 5. The effect of the Mandates becomes significant two years after the adoption of the
mandate. For this reason, our difference-in-difference strategy concentrates on the effects of
must-access mandates two years after the implementation of the program.
Table 2 presents the estimated effects of PDMPs on the number of children in foster care
by main reason of removal. We find no evidence of significant effects of PDMP on child
removals. However, mandatory PDMPs had significant effects on removals reducing both
cases associated with neglect and physical abuse (Table 3). The magnitude of the effect is
economically significant. The introduction of mandatory procedures reduced child removals
by 8%. Foster-care cases associated with neglect and physical abuse reduced by respectively
9% and 10%. These results imply a reduction of 467 removals per year or 0.27 less cases per
1000 children (Table A.2, Panel A and B). Table A.3 illustrates the sensitivity of our main
results to state-level time-varying controls and state specific time trends. While including
state-level controls reduces the magnitude of the coefficient, the coefficient remains negative
and statistically and economically significant. Using the method proposed by Oster (2017),
9
we estimate that to explain away our main result on child removals the extent of selection on
unobservables should be at least 14 times larger than the extent of selection on observables.
We don’t find significant effects among Blacks (Table A.4), while the effects are larger
among children aged 0-12 (Table A.5).
Our baseline findings suggest that must-access PDMPs may substantially reduce the costs
associated with child removals. With a back of the envelope calculation based on previous
estimates on the costs of child removals, our results imply that must-access PDMPs reduced
costs associated with child removals by approximately 476 million dollars per year, or 4.76
billion in 10 years.
5 Conclusion
The recent opioid epidemic has dramatic implications for the children of opioid-dependent
parents. As the opioid crisis spreads to urban counties and to different groups of the pop-
ulation, more children are at higher risk of neglect, abuse or removal from their parental
caregiver.
Our paper contributes to the literature on the effectiveness of drug monitoring programs.
We provide evidence that while operational PDMPs had no significant effects on foster
care caseloads, mandatory PDMP substantially reduced cases of child removals (-10%) with
significant reductions in the number of cases associated with neglect and physical abuse.
Given the evidence on the long-lasting effects of child maltreatment and foster care, our
findings suggest that the human capital, health and economic cost of the opioid crisis may
be very large. At the same point, the effectiveness of programs monitoring drug-prescription
supports the implementation of supply-side policies aimed at reducing the diffusion of opioid
substances in the population. The indirect effects of these policies on child well-being are not
negligible. Policy makers should take into account the human and financial costs of parental
drug abuse when evaluating policy effectiveness and allocating resources across programs
10
aimed at contrasting the opioid epidemic.
References
Alpert, A., Powell, D., Pacula, R. L., 2017. Supply-side drug policy in the presence of sub-
stitutes: Evidence from the introduction of abuse-deterrent opioids. Tech. rep., National
Bureau of Economic Research.
Borgschulte, M., Corredor-Waldron, A., Marshall, G., forth. A path out: Prescription drug
abuse and suicide. Journal of Economic Behavior and Organization.
Buchmueller, T. C., Carey, C., 2017. The effect of prescription drug monitoring programs
on opioid utilization in medicare. NBER Working Paper.
Case, A., Deaton, A., 2015. Rising morbidity and mortality in midlife among white non-
hispanic americans in the 21st century. Proceedings of the National Academy of Sciences
112 (49), 15078–15083.
Cunningham, S., Finlay, K., 2013. Parental substance use and foster care: Evidence from
two methamphetamine supply shocks. Economic Inquiry 51 (1), 764–782.
Currie, J., Spatz Widom, C., 2010. Long-term consequences of child abuse and neglect on
adult economic well-being. Child maltreatment 15 (2), 111–120.
Dart, R. C., Surratt, H. L., Cicero, T. J., Parrino, M. W., Severtson, S. G., Bucher-Bartelson,
B., Green, J. L., 2015. Trends in opioid analgesic abuse and mortality in the united states.
New England Journal of Medicine 372 (3), 241–248.
Dave, D. M., Grecu, A. M., Saffer, H., 2017. Mandatory access prescription drug monitoring
programs and prescription drug abuse. NBER Working Paper.
Doyle Jr, J. J., 2007. Child protection and child outcomes: Measuring the effects of foster
care. American Economic Review 97 (5), 1583–1610.
11
Doyle Jr, J. J., 2008. Child protection and adult crime: Using investigator assignment to
estimate causal effects of foster care. Journal of Political Economy 116 (4), 746–770.
Dube, S. R., Felitti, V. J., Dong, M., Chapman, D. P., Giles, W. H., Anda, R. F., 2003.
Childhood abuse, neglect, and household dysfunction and the risk of illicit drug use: the
adverse childhood experiences study. Pediatrics 111 (3), 564–572.
Kilby, A., 2015. Opioids for the masses: welfare tradeoffs in the regulation of narcotic pain
medications. Cambridge: Massachusetts Institute of Technology.
Laxmaiah Manchikanti, M., Standiford Helm, I., MA, J. W. J., PhD, V. P., MSc, J. S. G.,
DO, P., et al., 2012. Opioid epidemic in the united states. Pain physician 15, 2150–1149.
Mallatt, J., 2017. The effect of prescription drug monitoring programs on opioid prescriptions
and heroin crime rates.
Meara, E., Horwitz, J. R., Powell, W., McClelland, L., Zhou, W., O’malley, A. J., Morden,
N. E., 2016. State legal restrictions and prescription-opioid use among disabled adults.
New England Journal of Medicine 375 (1), 44–53.
Oster, E., 2017. Unobservable selection and coefficient stability: Theory and evidence. Jour-
nal of Business & Economic Statistics, 1–18.
Pacula, R. L., Powell, D., Taylor, E. A., 2015. Does Prescription Drug Coverage Increase
Opioid Abuse?: Evidence from Medicare. National Bureau of Economic Research.
Patrick, S. W., Fry, C. E., Jones, T. F., Buntin, M. B., 2016. Implementation of prescription
drug monitoring programs associated with reductions in opioid-related death rates. Health
Affairs 35 (7), 1324–1332.
Paulozzi, L. J., Mack, K. A., Hockenberry, J. M., 2014. Vital signs: variation among states
in prescribing of opioid pain relievers and benzodiazepines-united states, 2012. Morbidity
and Mortality Weekly Report 63 (26), 563–568.
12
Quast, T., Storch, E. A., Yampolskaya, S., 2018. Opioid prescription rates and child removals:
Evidence from florida. Health Affairs 37 (1), 134–139.
Rudd, R. A., 2016. Increases in drug and opioid-involved overdose deathsunited states, 2010–
2015. MMWR. Morbidity and mortality weekly report 65.
Ruhm, C. J., 2018. Deaths of despair or drug problems? Tech. rep., National Bureau of
Economic Research.
Simoni-Wastila, L., Qian, J., 2012. Influence of prescription monitoring programs on anal-
gesic utilization by an insured retiree population. Pharmacoepidemiology and drug safety
21 (12), 1261–1268.
Zill, N., 2011. Better prospects, lower cost: The case for increasing foster care adoption.
Adoption Advocate 35, 1–7.
13
Figures and Tables
Figure 1: Trends in Drug Related Deaths (2000-2015, CDC) and Drug-Related Foster CareCases
Notes - Data on drug-related deaths are drawn from CDC database on detailed mortality causes. The Underlying Cause of
Death database contains mortality and population counts for all U.S. counties. Data are based on death certificates for U.S.
residents. Each death certificate identifies a single underlying cause of death and demographic data. Data on children who were
assigned to foster care because of drug-related abuses are drawn from the Adoption and Foster Care Analysis and Reporting
System (AFCARS), Foster Care File (2000-2016).
14
Figure 2: Trends in Child Removals
Notes - Data on children who were assigned to foster care because of drug-related abuses are drawn from the Adoption and
Foster Care Analysis and Reporting System (AFCARS), Foster Care File (2000-2015).
15
Figure 3: Retail Drug Distribution by Drug Code for U.S.
Notes - Data are drawn from the Automated Reports and Consolidated Ordering System (ARCOS) provided by the U.S.
Department of Justice Drug Enforcement Administration, Diversion Control Division. Data cover the period 1999-2015.
16
Table 1: Summary Statistics, AFCARS (2000-2015)
Mean Standard deviation# removals 4519.496 5190.235
Reason:Neglect 3214.062 4151.742
Drug abuse 1432.445 1902.54Physical abuse 960.282 1313.524Alcohol abuse 419.026 539.051Sexual abuse 321.487 474.670
Notes - Data on children who were assigned to foster care because of drug-related abuses are drawn from the Adoption and
Foster Care Analysis and Reporting System (AFCARS), Foster Care File (2000-2015). The table reports unweighted summary
statistics by state and year for the main outcome variables in all US states. Data spans years 2000 to 2016.
17
Figure 4: PDMP Event Study-Child Removal
-.3-.2
-.10
.1C
hild
Rem
oval
s
PDMP
_3
PDMP
_2
PDMP
_1
PDMP
0
PDMP
1
PDMP
2
PDMP
3
-.3-.2
-.10
.1.2
Neg
lect
Cas
es
PDMP
_3
PDMP
_2
PDMP
_1
PDMP
0
PDMP
1
PDMP
2
PDMP
3
-.3-.2
-.10
.1Ph
ysic
al A
buse
s
PDMP
_3
PDMP
_2
PDMP
_1
PDMP
0
PDMP
1
PDMP
2
PDMP
3
Notes - All estimates include time-varying control at the state level for the share of female, Hispanic, African-American,
White, foreign-born, non-citizen population, average family income (log), unemployment rate, children population (0-18), year
and state fixed effects, state-specific time trends and the following laws/regulations: Good Samaritan laws, Doctor Shopping,
Pain Clinic regulations, Physician exams, require ID, and tamper-resistant PF. Data on children who were assigned to foster
care because of drug-related abuses are drawn from the Adoption and Foster Care Analysis and Reporting System (AFCARS),
Foster Care File (2000-2016).
18
Figure 5: Mandate Event Study-Child Removal
-.3-.2
-.10
.1C
hild
Rem
oval
s
Mand
ate_3
Mand
ate_2
Mand
ate_1
Mand
ate0
Mand
ate1
Mand
ate2
Mand
ate3
-.3-.2
-.10
.1.2
Neg
lect
Cas
es
Mand
ate_3
Mand
ate_2
Mand
ate_1
Mand
ate0
Mand
ate1
Mand
ate2
Mand
ate3
-.2-.1
0.1
.2.3
Phys
ical
Abu
ses
Mand
ate_3
Mand
ate_2
Mand
ate_1
Mand
ate0
Mand
ate1
Mand
ate2
Mand
ate3
Notes - All estimates include time-varying control at the state level for the share of female, Hispanic, African-American,
White, foreign-born, non-citizen population, average family income (log), unemployment rate, children population (0-18), year
and state fixed effects, state-specific time trends and the following laws/regulations: Good Samaritan laws, Doctor Shopping,
Pain Clinic regulations, Physician exams, require ID, and tamper-resistant PF. Data on children who were assigned to foster
care because of drug-related abuses are drawn from the Adoption and Foster Care Analysis and Reporting System (AFCARS),
Foster Care File (2000-2016).
19
Table 2: Effects of PDMP on Foster Cases (IHS)
(1) (2) (3)IHS abuse cases IHS neglect cases IHS physical abuses
PDMPt−2 -0.218 -0.205 -0.181(0.173) (0.176) (0.125)
Observations 867 867 867Mean of Dep. Var. 8.531 8.156 6.855Std.Dev. of Dep. Var. 1.448 1.422 1.400
Notes - All estimates include time-varying control at the state level for the share of female, Hispanic, African-American,
White, foreign-born, non-citizen population, average family income (log), unemployment rate, children population (0-18), year
and state fixed effects, state-specific time trends and the following laws/regulations: Good Samaritan laws, Doctor Shopping,
Pain Clinic regulations, Physician exams, require ID, and tamper-resistant prescription form requirement. Data on children
who were assigned to foster care because of drug-related abuses are drawn from the Adoption and Foster Care Analysis and
Reporting System (AFCARS), Foster Care File (2000-2016).
20
Table 3: Effects of Mandatory PDMP on Foster Cases (IHS)
(1) (2) (3)IHS abuse cases IHS neglect cases IHS physical abuses
Mandate t−2 -0.078** -0.100*** -0.089**(0.034) (0.034) (0.043)
Observations 371 371 371Mean of Dep. Var. 8.711 8.338 6.890Std.Dev. of Dep. Var. 0.908 0.967 0.973
Notes - All estimates include time-varying control at the state level for the share of female, Hispanic, African-American,
White, foreign-born, non-citizen population, average family income (log), unemployment rate, children population (0-18), year
and state fixed effects, state-specific time trends and the following laws/regulations: Good Samaritan laws, Doctor Shopping,
Pain Clinic regulations, Physician exams, require ID, and tamper-resistant prescription form requirement. Data on children
who were assigned to foster care because of drug-related abuses are drawn from the Adoption and Foster Care Analysis and
Reporting System (AFCARS), Foster Care File (2000-2016).
21
Appendix
22
Table A.1: Effects of PDMP on Foster Cases
(1) (2) (3)abuse cases neglect cases physical abuses
Number of cases
PDMPt−2 -348.367* -241.950 -125.614**(182.852) (178.239) (59.854)
Observations 867 867 867Mean of Dep. Var. 4543 3235 945.9Std.Dev. of Dep. Var. 5247 4231 1307
Cases per 1000 children
PDMPt−2 -0.017 0.005 -0.021(0.063) (0.058) (0.020)
Observations 867 867 867Mean of Dep. Var. 3.280 2.301 0.651Std.Dev. of Dep. Var. 1.473 1.170 0.482
Notes - All estimates include time-varying control at the state level for the share of female, Hispanic, African-American,
White, foreign-born, non-citizen population, average family income (log), unemployment rate, children population (0-18), year
and state fixed effects, state-specific time trends and the following laws/regulations: Good Samaritan laws, Doctor Shopping,
Pain Clinic regulations, Physician exams, require ID, and tamper-resistant PF. Data on children who were assigned to foster
care because of drug-related abuses are drawn from the Adoption and Foster Care Analysis and Reporting System (AFCARS),
Foster Care File (2000-2016).
23
Table A.2: Effects of Mandatory PDMP on Foster Cases
(1) (2) (3)abuse cases neglect cases physical abuses
Number of cases
Mandatet−2 -467.742** -394.359** -92.921***(181.931) (182.734) (27.222)
Observations 371 371 371Mean of Dep. Var. 4563 3407 775.8Std.Dev. of Dep. Var. 5105 4517 982
Cases per 1000 children
Mandatet−2 -0.276** -0.202** -0.049***(0.125) (0.081) (0.016)
Observations 371 371 371Mean of Dep. Var. 3.556 2.541 0.598Std.Dev. of Dep. Var. 1.554 1.306 0.317
Notes - All estimates include time-varying control at the state level for the share of female, Hispanic, African-American,
White, foreign-born, non-citizen population, average family income (log), unemployment rate, children population (0-18), year
and state fixed effects, state-specific time trends and the following laws/regulations: Good Samaritan laws, Doctor Shopping,
Pain Clinic regulations, Physician exams, require ID, and tamper-resistant PF. Data on children who were assigned to foster
care because of drug-related abuses are drawn from the Adoption and Foster Care Analysis and Reporting System (AFCARS),
Foster Care File (2000-2016).
24
Table A.3: Must-access PDMP and Foster Care Admissions
(1) (2) (3) (4)
Child removals
Mandatet−2 -587.285* -446.528 -611.210** -463.499**(303.077) (275.098) (290.749) (183.450)
Observations 371 371 371 371Mean of Dep. Var. 4563 4563 4563 4563Std.Dev. of Dep. Var. 5105 5105 5105 5105
Child removal cases associated with neglect
Mandatet−2 -850.913** -509.274* -616.379* -390.716**(341.638) (293.915) (313.831) (184.450)
Observations 371 371 371 371Mean of Dep. Var. 3407 3407 3407 3407Std.Dev. of Dep. Var. 4517 4517 4517 4517
Child removal cases associated withPhysical Abuse
Mandatet−2 106.755 -15.254 -31.471 -92.894***(92.016) (53.570) (71.080) (27.333)
Observations 371 371 371 371Mean of Dep. Var. 775.8 775.8 775.8 775.8Std.Dev. of Dep. Var. 982 982 982 982
State F.E. YES YES YES YESYear F.E. YES YES YES YESTime-varying state-levelc controls NO YES YES YESOther laws NO NO YES YESState specific time trends NO NO NO YES
25
Table A.4: Effects of Mandatory PDMP on Foster Cases (Races)
(1) (2) (3)IHS abuse cases IHS neglect cases IHS physical abuses
Whites
Mandatet−2 -0.040 -0.057* -0.056(0.039) (0.030) (0.050)
Observations 371 371 371Mean of Dep. Var. 9.060 8.706 7.175Std.Dev. of Dep. Var. 0.911 0.962 0.960
Blacks
Mandatet−2 -0.024 -0.031 -0.086(0.055) (0.053) (0.067)
Observations 371 371 371Mean of Dep. Var. 7.726 7.396 6.038Std.Dev. of Dep. Var. 1.602 1.621 1.690
Notes - All estimates include time-varying control at the state level for the share of female, Hispanic, African-American,
White, foreign-born, non-citizen population, average family income (log), unemployment rate, children population (0-18), year
and state fixed effects, state-specific time trends and the following laws/regulations: Good Samaritan laws, Doctor Shopping,
Pain Clinic regulations, Physician exams, require ID, and tamper-resistant PF. Data on children who were assigned to foster
care because of drug-related abuses are drawn from the Adoption and Foster Care Analysis and Reporting System (AFCARS),
Foster Care File (2000-2016).
26
Table A.5: Effects of Mandatory PDMP on Foster Cases (Age groups)
(1) (2) (3)IHS abuse cases IHS neglect cases IHS physical abuses
Age 0-6
Mandatet−2 -0.078** -0.099*** -0.096*(0.038) (0.035) (0.051)
Observations 371 371 371R-squared 0.996 0.994 0.993Mean of Dep. Var. 8.169 7.810 6.218Std.Dev. of Dep. Var. 0.926 0.974 1.026
Age 7-12
Mandatet−2 -0.093** -0.115*** -0.071(0.038) (0.039) (0.043)
Observations 371 371 371Mean of Dep. Var. 7.296 6.920 5.559Std.Dev. of Dep. Var. 0.912 0.972 0.968
Age 13-18
Mandatet−2 -0.068* -0.089* -0.119*(0.036) (0.048) (0.061)
Observations 371 371 371Mean of Dep. Var. 6.928 6.500 5.324Std.Dev. of Dep. Var. 0.904 1.006 0.941
Notes - All estimates include time-varying control at the state level for the share of female, Hispanic, African-American,
White, foreign-born, non-citizen population, average family income (log), unemployment rate, children population (0-18), year
and state fixed effects, state-specific time trends and the following laws/regulations: Good Samaritan laws, Doctor Shopping,
Pain Clinic regulations, Physician exams, require ID, and tamper-resistant PF. Data on children who were assigned to foster
care because of drug-related abuses are drawn from the Adoption and Foster Care Analysis and Reporting System (AFCARS),
Foster Care File (2000-2016).
27
IntroductionBackground and DataPrescription Drug Epidemic and Policy ResponseData
Empirical SpecificationMain ResultsTrends and Descriptive StatisticsPDMPs and Foster Care Caseloads
Conclusion